Merge pull request 'Initial version of atoi' (#15) from PleegWat/symphony_stdlib:str_to_int into main

Reviewed-on: #15

src/stdlib.asm

@@ -3,6 +3,7 @@ pub include imath
 pub include array
 pub include console
 pub include mem
+pub include string_to_int
 
 ; Needs to be last!
 pub include LUTs

src/string_to_int.asm

@@ -0,0 +1,188 @@
+; Internal register assignments:
+; r1 - Partially parsed output integer
+; r2 - Parsing position
+; r3 - Character being parsed
+; r4 - Set to -1 if the input is negative
+
+; Convert string to integer (akin to libc atoi(), strotoi())
+; Arguments:
+; r1 - Pointer to string
+; Result:
+; r1 - Parsed integer
+; r2 - Pointer to first rejected input byte
+; Clobbers:
+; flags
+; r3 - last character read
+; r4 - Negative marker
+; Note: Unless the input is "0", tail-calls into a base-specific specialization.
+pub auto:
+mov r2, r1
+mov r1, 0
+mov r4, 0
+load_8 r3, [r2]
+cmp r3, 0x2D ; '-'
+jne auto_positive
+sub r4, zr, 1           ; Set r4 to -1
+add r2, r2, 1
+auto_positive:
+load_16 r3, [r2]        ; 2-byte prefix "0b", "0o", "0x", etc.
+or r3, r3, 0x20         ; 2nd char to lower case
+add r2, r2, 2
+cmp r3, 0x3062 ; "0b"
+je bin_loop
+cmp r3, 0x306F ; "0o"
+je oct_loop
+cmp r3, 0x3078 ; "0x"
+je hex_loop
+sub r2, r2, 2           ; no matching prefix, move pointer back
+jmp dec_loop
+
+; Epilogue
+done:
+add r1, r1, r4          ; If r4 is -1, negate r1. Else it's 0 and no effect.
+xor r1, r1, r4
+mov flags, 0            ; No error
+jmp r13
+
+; Convert decimal string to integer
+; Arguments:
+; r1 - Pointer to string
+; Result:
+; r1 - Parsed integer
+; r2 - Pointer to first rejected input byte
+; Clobbers:
+; flags
+; r3 - last character read
+; r4 - Negative marker
+pub dec:
+mov r2, r1
+mov r1, 0
+mov r4, 0
+load_8 r3, [r2]
+cmp r3, 0x2D ; '-'
+jne dec_positive
+sub r4, zr, 1           ; Set r4 to -1
+add r2, r2, 1
+dec_loop:
+load_8 r3, [r2]
+dec_positive:
+sub r3, r3, 0x30 ; '0'
+cmp r3, 9
+ja done
+add r2, r2, 1
+lsl flags, r1, 2        ; Use flags to help multiply by 10
+add r1, r1, flags
+lsl r1, r1, 1
+add r1, r1, r3
+jmp dec_loop
+
+; Convert binary string to integer
+; Arguments:
+; r1 - Pointer to string
+; Result:
+; r1 - Parsed integer
+; r2 - Pointer to first rejected input byte
+; Clobbers:
+; flags
+; r3 - last character read
+; r4 - Negative marker
+pub bin:
+mov r2, r1
+mov r1, 0
+mov r4, 0
+load_8 r3, [r2]
+cmp r3, 0x2D ; '-'
+jne bin_positive
+sub r4, zr, 1           ; Set r4 to -1
+add r2, r2, 1
+bin_positive:
+load_16 r3, [r2]        ; check for prefix
+or r3, r3, 0x20         ; 2nd char to lower case
+cmp r3, 0x3062 ; "0b"
+jne bin_loop
+add r2, r2, 2
+bin_loop:
+load_8 r3, [r2]
+sub r3, r3, 0x30 ; '0'
+cmp r3, 1
+ja done
+add r2, r2, 1
+lsl r1, r1, 1
+add r1, r1, r3
+jmp bin_loop
+
+; Convert octal string to integer
+; Arguments:
+; r1 - Pointer to string
+; Result:
+; r1 - Parsed integer
+; r2 - Pointer to first rejected input byte
+; Clobbers:
+; flags
+; r3 - last character read
+; r4 - Negative marker
+pub oct:
+mov r2, r1
+mov r1, 0
+mov r4, 0
+load_8 r3, [r2]
+cmp r3, 0x2D ; '-'
+jne oct_positive
+sub r4, zr, 1           ; Set r4 to -1
+add r2, r2, 1
+oct_positive:
+load_16 r3, [r2]        ; check for prefix
+or r3, r3, 0x20         ; 2nd char to lower case
+cmp r3, 0x306F ; "0o"
+jne oct_loop
+add r2, r2, 2
+oct_loop:
+load_8 r3, [r2]
+sub r3, r3, 0x30 ; '0'
+cmp r3, 7
+ja done
+add r2, r2, 1
+lsl r1, r1, 3
+add r1, r1, r3
+jmp oct_loop
+
+; Convert hexadecimal string to integer
+; Arguments:
+; r1 - Pointer to string
+; Result:
+; r1 - Parsed integer
+; r2 - Pointer to first rejected input byte
+; Clobbers:
+; flags
+; r3 - last character read
+; r4 - Negative marker
+pub hex:
+mov r2, r1
+mov r1, 0
+mov r4, 0
+load_8 r3, [r2]
+cmp r3, 0x2D ; '-'
+jne hex_positive
+sub r4, zr, 1           ; Set r4 to -1
+add r2, r2, 1
+hex_positive:
+load_16 r3, [r2]        ; check for prefix
+or r3, r3, 0x20         ; 2nd char to lower case
+cmp r3, 0x3078 ; "0x"
+jne hex_loop
+add r2, r2, 2
+hex_loop:
+load_8 r3, [r2]
+sub r3, r3, 0x30 ; '0'
+cmp r3, 10
+jb hex_add
+sub r3, r3, 0x11 ; 'A' - '0'
+and r3, r3, 0xdf        ; to lower case
+cmp r3, 5               ; 0-5: 6 letters
+ja done
+add r3, r3, 10          ; Adjust for digits below
+hex_add:
+add r2, r2, 1
+lsl r1, r1, 4
+add r1, r1, r3
+jmp hex_loop

tests/string_to_int_test.asm

@@ -0,0 +1,57 @@
+; test harness for string_to_int
+; Registers:
+; r8 - test address
+; r9 - expected result
+; r10 - actual length
+; r11 - expected length
+li r8, tests
+li r12, end_of_tests
+next_test:
+; Set up arguments and run test
+add r1, r8, 5          ; Start of test string
+mov r13, atoi_test_ret
+jmp stdlib.string_to_int.auto
+atoi_test_ret:
+
+; Load reference data
+load_32 r9, [r8]       ; Expected result
+sub r10, r2, r8
+sub r10, r10, 5        ; Actual string length
+add r11, r8, 4         ; Address of expected length
+load_8 r11, [r11]      ; Expected length
+
+; Verify results
+cmp r1, r9
+incorrect_result: jne incorrect_result
+cmp r10, r11
+incorrect_length: jne incorrect_length
+
+; Next test
+add r8, r8, 0x10
+cmp r8, r12
+jl next_test
+
+; Done
+success: jmp success
+
+include ../src/stdlib
+
+@0x20000
+tests:
+; inlen is the number of input bytes the function is expected to consume
+; addr  result          inlen  instr
+@0x20000 U32 0           U8 0   "\0"
+@0x20010 U32 0           U8 1   "0\0"
+@0x20020 U32 1           U8 1   "1\0"
+@0x20030 U32 2           U8 1   "2:\0"
+@0x20040 U32 42          U8 2   "42\0"
+@0x20050 U32 67          U8 2   "67lol\0"
+@0x20060 U32 0x69a       U8 5   "0x69a@\0"
+@0x20070 U32 0x4B4       U8 5   "0x4B4g\0"
+@0x20080 U32 0o23        U8 4   "0o239\0"
+@0x20090 U32 0b1011      U8 6   "0b1011\0"
+@0x200a0 U32 0b10001     U8 7   "0b100012\0"
+@0x200b0 U32 0xFFFFFFEB  U8 3   "-21\0"       ; Yuk
+@0x200c0
+end_of_tests:
+